Zenodo
Sang, Zihaohan;
Robitaille, Alec L.;
Stralberg, Diana
—
2025-01-10
Increasing frequency and severity of drought events associated with global warming present a key challenge for forest ecosystems. Tree growth reductions in response to climatic dryness may be used to measure the drought sensitivity of forest ecosystems. Investigations of the environmental conditions that determine drought sensitivity are needed to accurately predict ecosystem responses under projected climate change. Tree-ring records have been used to assess drought-related growth responses for dominant tree species. However, the application of dendroecological approaches across large regions is limited by the spatial and temporal availability of tree ring records.
In this study, we modified the approach of Cartwright et al. (2020) to develop an index of drought sensitivity across Canadian forested regions. We used high-resolution satellite products (LANDSAT 5, 7, 8, 30-m resolution) to quantify forest responses under different durations of water stress (3-month, 12-month, and 3-year, representing short-, mid-, and long-term dryness, respectively). We analyzed normalized difference vegetation (NDVI) index data during the peak of growing seasons from 1985 to 2020. The deviation in NDVI between normal years and dry years was used to represent local drought sensitivity. We then identified vegetation and landscape characteristics that are associated with drought sensitivity and generated spatial predictions of drought sensitivity (and the inverse, refugia from drought) across the study area.
We found that forests located in low-latitude, central boreal, and interior British Columbia regions were limited by dry climate conditions, especially for short- and intermediate-term drought events. Sensitivity to mid- and long-term drought was most common in high-latitude regions. Also, high-elevation sites were generally less sensitive than lower elevations in the same region. Site climate conditions and soil factors were found to be stronger drivers of drought sensitivity than topographic characteristics. Our high-resolution assessment can be used to identify which forests have exhibited the highest and lowest sensitivity to drought over the last four decades, informing regional land-use planning priorities and facilitating local management interventions to mitigate and adapt to climate change.
This data consists of 3 GeoTIFF rasters representing drought sensitivity predictions for 3-month, 12-month, and 3-years across Canadian boreal forest ecosystems (‘drought_pred_3mo.tif’, ‘drought_pred_12mo.tif’, ‘drought_pred_3yr.tif). All three rasters have ~1-km resolution (0.0089 degree), using EPSG:4326 as coordinate reference system. The original inputs to drought sensitivity models were at 100-m resolution. To reduce the output size, we resampled these rasters with nearest neighbor and exported 1-km resolution files for sharing purposes.
Positive values indicate high predicted drought sensitivity (severe growth reduction in drier years), while negative values represent potential benefits from warmer, water-limited periods. Unvegetated areas are excluded from models and predictions An Earth Engine app for visualizing drought sensitivity is available at https://ee-sang4gee.projects.earthengine.app/view/droughtindex
https://creativecommons.org/licenses/by/4.0/legalcode